[Bug] Context window calculation mismatch for 200K model

(Another user.) The 1M window isn't a property the model carries automatically — it's gated on two separate things, and the 200K you're seeing is the fallback when either one is missing:

1. The model variant. The 1M-context capability is a distinct variant (the [1m] / 1M-context build), not just "the same model with a bigger number." If the selected model isn't the 1M variant, the effective cap is 200K. Check /model and confirm you've actually got the 1M variant selected, not the base one.
2. The entitlement. Even with the 1M variant, the 1M window is access-gated (it rolled out behind a beta + higher-tier entitlement, not on by default for every account). If your account/tier isn't entitled to 1M context, Claude Code falls back to the standard 200K — which is exactly the cap your screenshot shows.

So the 200K display is usually correct behavior for one of those two gaps, not a calculation bug. To narrow it down:

• Confirm via /model that the 1M variant is the one in use.
• Confirm your plan/tier actually includes 1M context (it's not universal — API access tiers and subscription tiers differ here).

If both are satisfied and it still caps at 200K — that's the real bug worth reporting, and the sharp framing for Anthropic is: 1M variant selected + account entitled, yet the window is computed as 200K on 2.1.190 specifically. Add whether it showed 1M on an earlier version (your "Is this a regression?" answer) — if it regressed at a specific build, that points them straight at the context-size computation change between those versions. A screenshot of /model alongside the 200K readout makes it unambiguous.

Switching models didn't help. Had to completely restart and now works in v2.1.193.

This issue has been automatically locked since it was closed and has not had any activity for 7 days. If you're experiencing a similar issue, please file a new issue and reference this one if it's relevant.